Life of a Lab Rat

Gee, I'd love to be of assistance, but I'm too busy staring blankly at my cool algorithm/fast computer/nifty widget. Do you always start a request for help by insulting the very people who are among the most likely to be able to help you?

I'm not sure I understand your problem. Why are you reading them into an array of arrays if your final goal is to get them into four arrays? Why can't you just do
while (>)
{
my ($probeid, $cn, $start, $stop) = split ",";
push @probeids, $probeid;
push @cns, $cn;
...
}

I do enjoy pushing hot buttons.

Paul, I didn't think of that. Thanks. I was stuck in a readline/accumulator nightmare.

One thing that annoys me about biologists is their unwillingness to learn how to wrangle code to good effect ;) There are plenty of biology-trained bioinformaticians, able and willing to help (I'm one). Most often, nobody asks or if they do, they ask badly.

I often find that Perl complex data structures (hashes of hashes, hashes of arrays...) are the way to store biological data. You have a chromosome; on that chromosome is a probe; the probe has a position. The trick is to map that description into a data structure. The unique identifiers - chromosome and probe - would make good hash keys. The start/end makes a good array. Result - a hash of hashes of arrays. Something like:

@{$hash{$cn}{$probeid}} = ($start, $end)

Probes matching in more than one location? No problem; push [$start, $end] instead, giving a hash (key chromosome) of hashes (key probeids) of arrays (number of positions) of arrays (position starts/ends).

Precisely Neil - I believe you started out as a biologist so you understand the problem (from both sides). Paul (check out his website) is a smart computer cookie, who doesn't pretend to know anything biological, so again, someone I'd be happy to go to for help. It's the "ooh, no, why don't you do this because it's more interesting?" attitude that bugs me - and that I've come up against a lot.

Which means I often find it less hassle in the long term to roll up my sleeves and try to hack out something. Naturally, I was cautious to use 'most' in my first paragraph precisely so that I could filter out the unhelpful ones. . .

You've given me something to think about, Neil. I might step back and have a serious think about the workflow. The lab will be quiet next week - Combio :)

Yeah, I'm exposed regularly to both sides of the coin. Bioinformatics should be about solving biological problems and far too much of it is not - see any issue of the journal Bioinformatics, for example! On the other hand, a lot of biologists are very poor at engaging with bioinformaticians, for a multitude of reasons.

We need far more people who are comfortable speaking "biology" and "programming", since bioinformatics is really just the process of converting back and forth between biological "objects" and their representation as code. In many cases, there's a direct mapping (sequence - string of characters, for a trivial example).

I'll be manning POS-MON-86, 1:30-2:30 next Monday.

Neil wrote: "if they do, they ask badly" and I fully agree from the biologist side of things. We don't usually have the words to ask the questions right!

In the chronic shortage of bioinformaticians in my institution, we biologists had to use the few tools we could manage - basics of Excel and Access (sorry about FMPro, but we're slaves to Microsoft and untrained in programming) to do something that any proper programmer could probably have done in one tenth the time and with far more elegance and efficiency. This was similar to BK's task - take a long list of gene IDs that are interesting (SAGE data this time), map the physical interval of the whole gene as known; on the other side, convert the list of intervals corresponding to human disease loci, available from OMIM, into physical coordinates, and find out which of our interesting genes fall within the known disease intervals (with a margin).

I am describing this task as a biologist would, see?
What would be great is to have some sort of list on OpenWetWare of interested bioinformaticians around the world who say, sure, I'm available and would be happy to collaborate (making sure it is considered a true collaboration and not just the usual use-the-tech attitude). Anyone?